Thursday, February 04, 2010

The Salience Principle

Here is a little nugget of trading wisdom: the market systematically punishes salience. Show me a strategy that makes use of highly salient information (i.e., information that is likely to stick in the mind at first, casual exposure) and I'll show you a strategy that underperforms.

Many mechanical/algorithmic programs make use of this principle, particularly at short time frames.

Going back to 2002, if you bought the S&P 500 Index (SPY) following an up day during an up week, the next week averaged a price loss of -.18% (338 up, 324 down). If you bought the S&P 500 Index following a down day during a down week, the next week averaged a price gain of .33% (283 up, 201 down).

Market technician Joe Granville famously asserted that if it's obvious, it's obviously wrong.

That's the salience principle, and it's why impulsive trades so often are losers.



Radek Dobias, H.B.Sc., M.W.S., B.Ed. said...

One must know one's market the way one knows a loved one: you look through the surface and see what's underneath.

David Merkel said...

That is why I am reluctant to insist that inflation and interest rates are going higher. Too many believe it. Too many are betting on it.

All that said, it may be more accurate to say that things don't happen that people are relying on not to happen. Mere opinion does not require taking an investment position. Crowded trades have a lot of people who believe the same thing, but even more, they are relying on it.

An Honest Republican said...

Wouldn't a gain following a loss and a loss following a gain be more like "the regression principle?"

stevenstevo said...

Huh? So I'm assuming the assumption here is that is an obvious winning strategy? Which if I am reading correctly is to buy the S&P 500 on a day following an up day during an up week and then sell at the end of the next week?

Never heard of that obvious strategy, and I'm not even sure it could be one.

For starters, it's impossible to know when there is going to be an "up week" or a "down week" until the week is already over.

And what if the return from the remaining days in the "up week" exceed the decline in the return for the following week?

And further, a positive return following a negative return can be misleading. Say you buy the stock in a down week at $100 and it then drops 40% to $60 by the end of the current down week. The next week it goes up 50% and you sell, which seemingly indicates a good trade. However, the price ends up at $90, so you actually lost 10% on the trade.